Scientists from the Yuri Gagarin Saratov State Technical University (SSTU), together with their foreign colleagues, have created a neural network capable of doubting at the time of decision-making, - writes RIA Novosti with reference to the journal of the American Institute of Physics.
By simulating the state of uncertainty of choice at the moment the brain makes a decision, the developers planned to capture a certain state in the general activity of the brain. It was rather difficult to recognize and describe the uncertainty at the moment a person made a decision.
To do this, the scientists analyzed the signals of magnetoencephalograms - recordings of the magnetic activity of the neural ensembles of the brain using artificial intelligence and machine learning methods.
The brain signals of each of the volunteers were recorded in a small experiment. People were shown optical illusions. People had to concentrate on the images in order to interpret them in some way. Neural networks were engaged in the analysis of the data obtained, which learned to recognize and describe in detail the states of uncertainty of human choice at the moment the brain makes a decision.
“Moreover, we were able to accurately measure the period of time during which a person hesitates and cannot make a choice,” said Professor Alexander Khramov, one of the authors of the study.
It was rather difficult to identify such states of brain activity using traditional methods, but using artificial intelligence for this, scientists were able to achieve success.
A neural network trained to "doubt" can be useful for scientists to create artificial intelligence capable of doubting the correctness of a decision. Before giving a final answer, such an AI will stop for a minute to think carefully. Who knows, maybe eventually robots from the future will decide not to kill “all humans” precisely because of this international scientific work?
Viacheslav Larionov
Promotional video: